Longitudinal patent analysis for nanoscale science

JournalofNanoparticleResearch5:333–363,2003.©2003KluwerAcademicPublishers.PrintedintheNetherlands.Longitudinalpatentanalysisfornanoscalescienceandengineering:Country,institutionandtechnologyfieldZanHuang1,HsinchunChen1,AlanYip1,GavinNg1,FeiGuo1,Zhi-KaiChen1andMihailC.Roco21ArtificialIntelligenceLab,DepartmentofManagementInformationSystems,EllerCollegeofBusinessandPublicAdministration,TheUniversityofArizona,Tucson,AZ85721,USA(E-mail:[e-mail]);2NationalScienceFoundation,4201WilsonBlvd.,Arlington,VA22230,USA(E-mail:[e-mail])Received3April2003;acceptedinrevisedform25April2003Keywords:patentanalysis,nanotechnology,nanoscience,knowledgediscovery,informationvisualization,self-organizingmap,citationnetworksAbstractNanoscalescienceandengineering(NSE)andrelatedareashaveseenrapidgrowthinrecentyears.Thespeedandscopeofdevelopmentinthefieldhavemadeitessentialforresearcherstobeinformedontheprogressacrossdifferentlaboratories,companies,industriesandcountries.Inthisproject,weexperimentedwithseveralanalysisandvisualizationtechniquesonNSE-relatedUnitedStatespatentdocumentstosupportvariousknowledgetasks.Thispaperpresentsresultsonthebasicanalysisofnanotechnologypatentsbetween1976and2002,contentmapanalysisandcitationnetworkanalysis.Thedatahavebeenobtainedonindividualcountries,institutionsandtech-nologyfields.Thetop10countrieswiththelargestnumberofnanotechnologypatentsaretheUnitedStates,Japan,France,theUnitedKingdom,Taiwan,Korea,theNetherlands,Switzerland,ItalyandAustralia.Thefastestgrowthinthelast5yearshasbeeninchemicalandpharmaceuticalfields,followedbysemiconductordevices.Theresultsdemonstratepotentialofinformation-baseddiscoveryandvisualizationtechnologiestocaptureknowledgeregard-ingnanotechnologyperformance,transferofknowledgeandtrendsofdevelopmentthroughanalyzingthepatentdocuments.IntroductionRecentrapiddevelopmentofNanoscaleScienceandEngineering(NSE)promisestobringfundamentalchangestoawiderangeofresearchfieldsandindus-tries.NSEhasbeenrecognizedtobecriticaltoacountry’sfuturescienceandtechnologycompetenceandhasrecentlyattractedglobalresearchanddevel-opmentinterests.TheUnitedStateshasannouncedtheNationalNanotechnologyInitiative(NNI)in2000basedonalong-termvision(Rocoetal.,2000).Morethan30countrieshaveadoptednationalprojectsorprogramspartiallystimulatedbyNNI(Roco,2002).Bothlong-termbasicresearchandshort-termdevel-opmentrelatedtonanotechnologyarebeingactivelyexploredacrossmanyscientificfieldsandindustrialapplications.Thespeedandscopeofnanotechnologydevelopmentmakeitcriticalforresearcherstobeawareofpublicationsandpatentsinthefieldacrossdifferentlaboratories,companies,industriesandcountries.Apatentisaspecialtypeoftechnologydocument.Asanopensourceofknowledge,itcontainsrichcon-tentregardingtechnologyinnovationsandisaccessiblebythegeneralpublic.Mostcountrieshaveadoptedsimilarpatentsystems.Alargenumberofpatentsareissuedeverydayandcollectedandpublishedsys-tematicallyworldwide.Forexample,USPatentandTrademarkOffice(USPTO)hasintotalmorethan5millionpatents,with3500to4000newlygrantedpatentsbeingaddedintothedatabaseeachweek.As334aresult,collectionsoffull-textpatentsoveralongperiodoftime(typically20–30years)areavailable.Thepatentdocumentsarealsostrictlystructured,pro-vidingstandardizedfieldssuchaspatentcitation,issuedate,assignee(theinstitutiontowhichthepatentisassignedto),inventors,technologyfieldclassification,andcountryandcityoftheassigneeandinventors,etc.Allthesespecialfeaturesofpatentdocumentsmakethemavaluablesourceofknowledgeregardingtechnologydevelopment.Weaimtoleveragevariousinformationanalysisandvisualizationtechnologiestosupportdomain-specificknowledgediscoveryfrompatentdocuments.Theproposedframeworkistargetedatsupportinghigh-levelknowledgetasks(e.g.country-leveltechnologystrengthcomparison,newresearchfieldidentification,etc.).Suchknowledgeistypicallyobtainedbyexten-siveliteraturesearchesthatrequirelargeamountsoftime,resourcesandhumanefforts.AnautomaticpatentanalysisframeworkhasthepotentialtoalleviatetheinformationoverloadproblemfacedbytheresearchersintheNSEfield.Thereisasubstantialacademicliteratureandmanyindustrialpracticesofusingpatentanalysisfortech-nologystrengthandtrendevaluation(Garfield,1955;Karki,1997;Oppenheim,2000).However,buildinganautomaticpatentanalysisservicefortheNSEisstillachallengingtask.Thedifficultiesare:(1)uncertaintyofthevalidityofusingpatentdatatoapproximatethescienceandtechnologydevelopmentinNSE;(2)diffi-cultyofintuitivepresentationofanalysisresults,suchasidentificationoffast-evolving(obsolescing)subcat-egoriesandnewconcepts;and(3)terminology/namingdifferencesthatareinherentinthepatentdata.Ourgoalistobuildaprototypesystemtoexaminebothtechni-calissuesandfundamentalhypothesesinvolvedwithknowledgediscoverythroughpatentanalysis.AsmallerscalesurveyoftheUSPTOdatabasethatsurveyed2600patentswasrunbyM.Meyer(seeRoco&Bainbridge,2001,pp.296–311).Thereviewedpatentshadthedominantfocusoninstruments,elec-tronicsandchemical/pharmaceutics.Anotherobserva-tionwasthatonlyabout1%ofthemwerereferredintheScienceCitationIndexpublicatio